Skip to content

karenina verify-status

Inspect a progressive save state file and show job status.

karenina verify-status [OPTIONS] STATE_FILE

The verify-status command shows the progress of a verification job that was started with --progressive-save. It reads the .state sidecar file and displays completed and pending tasks, elapsed time, model configuration, and file status.

Arguments

Argument Type Required Description
STATE_FILE PATH Yes Path to the .state file to inspect

Options

Option Short Type Default Description
--show-tasks -t flag False Show detailed list of completed and pending task IDs
--show-questions -q flag False Show list of completed and pending question IDs
--json -j flag False Output status as JSON instead of formatted text

Default Output

Without any flags, the command displays a Rich-formatted summary with five sections:

  1. Header — "Job Complete" (green) or "Job In Progress" (yellow)
  2. Progress bar — Visual completion indicator with percentage and counts
  3. Summary table — State file path, output path, benchmark path, completed/pending/total counts, timing
  4. Configuration table — Answering models, parsing models, replicate count
  5. Files table — Whether .tmp and .state sidecar files exist, with file sizes

If the job is incomplete, a resume hint is shown:

To resume: karenina verify --resume results.json.state

JSON Output

With --json, the command outputs a single JSON object suitable for scripting:

{
  "state_file": "results.json.state",
  "output_path": "results.json",
  "benchmark_path": "checkpoint.jsonld",
  "progress": {
    "total_tasks": 200,
    "completed_count": 90,
    "pending_count": 110,
    "progress_percent": 45.0
  },
  "timing": {
    "created_at": "2026-01-15T10:30:00",
    "last_updated_at": "2026-01-15T11:15:42",
    "elapsed_seconds": 2742.5
  },
  "config": {
    "answering_models": ["answering-1 (claude-haiku-4-5)"],
    "parsing_models": ["parsing-1 (claude-haiku-4-5)"],
    "replicate_count": 1
  },
  "files": {
    "tmp_file_exists": true,
    "tmp_file_size": 524288
  },
  "completed_question_ids": ["urn:uuid:abc123", "urn:uuid:def456"],
  "pending_question_ids": ["urn:uuid:ghi789"],
  "completed_task_ids": ["urn:uuid:abc123\tanswering-1\tparsing-1"],
  "pending_task_ids": ["urn:uuid:ghi789\tanswering-1\tparsing-1"]
}

Question IDs vs Task IDs

A question ID identifies a single benchmark question (URN format). A task ID identifies a specific verification task — the combination of a question, answering model, parsing model, and optional replicate number:

{question_id}\t{answering_canonical_key}\t{parsing_canonical_key}[\trep{N}]

For a benchmark with 10 questions, 2 answering models, and 3 replicates, there are 10 unique question IDs but 60 task IDs (10 x 2 x 3).

Use --show-questions for a high-level view by question. Use --show-tasks for the full task-level breakdown.

State File Format

The .state file is a JSON file created by karenina verify --progressive-save. It contains:

Field Type Description
format_version string State file format version ("1.0")
created_at string ISO 8601 timestamp when the job started
last_updated_at string ISO 8601 timestamp of the last completed task
benchmark_path string Path to the benchmark checkpoint
output_path string Path to the output file
config_hash string MD5 hash of the verification config
config object Full VerificationConfig as JSON
task_manifest array All task IDs for the job
completed_task_ids array Task IDs that have finished
total_tasks integer Total number of tasks
completed_count integer Number of completed tasks
start_time float Unix timestamp when the job started

The .state file is created alongside the output file (e.g., results.json.state for results.json).

Exit Codes

Code Meaning
0 State file inspected successfully
1 Error — state file not found or invalid format

Examples

Basic status check

karenina verify-status results.json.state

Show question-level progress

karenina verify-status results.json.state --show-questions

Show all task IDs (verbose)

karenina verify-status results.json.state --show-tasks

JSON output for scripting

karenina verify-status results.json.state --json

Pipe JSON to jq for filtering

karenina verify-status results.json.state --json | jq '.progress'

Full progressive save workflow

# 1. Start verification with progressive save
karenina verify checkpoint.jsonld --preset default.json \
  --output results.json --progressive-save

# 2. If interrupted, check status
karenina verify-status results.json.state

# 3. Resume from where it left off
karenina verify --resume results.json.state